[LinuxBIOS] [PATCH] Via CN700
rminnich at gmail.com
Fri Nov 16 23:25:46 CET 2007
Corey, your responses to these answers are good and I wonder if they
might go as comments in the code.
In general, if a register setting is so confidential we can't talk
about it, we could just
tag the block with
/* NDA required */
/* Done NDA required */
or some such couldn't we?
On Nov 16, 2007 12:19 AM, Corey Osgood <corey.osgood at gmail.com> wrote:
> New patch attached, I think I've cleared everything up that can be.
> Board patch will have to wait until tomorrow, I need to be up in <5 hours.
> Uwe Hermann wrote:
> > On Thu, Nov 15, 2007 at 12:38:01PM -0500, Corey Osgood wrote:
> > Pretty scary for someone without the datasheet. Please add more comments
> > and/or use self-explanatory #defines or something alike.
> For the most part these shouldn't be messed with, the values are from
> the porting guide and should work fine with any board (YMMV though).
> I've added some comments, but to do any real work on this would
> absolutely require access to the datasheets.
So just tag them as magic values and we are fine.
> And since the datasheets are NDA'd, completely reproducing them in
> comments could cause some problems. Most of these registers have
> different functions for every bit, so this file could easily become very
> large very fast.
Ollie once pointed out to me that we don't want to replicate the data
book in the code, I agree with you on this. Past a point, people can
go to the data book.
> It's the most messed up form of interleaving I've ever heard of. It
> allows the northbridge to trick /itself/ into thinking 2 dimms are a
> single dimm, via a virtual bank routine and this register. I can't come
> up with any useful purpose, since the number of virtual and physical
> banks supported is the same. You can also (heh) map any physical bank
> (or combination of physical banks) to any virtual bank, to really mess
> with someone's head. It makes me wonder if Via is taking bribes from the
> makers of Excedrin or Heineken.
Maybe without the more extreme comments, this might be useful in the code :-)
Anyway, I'm looking forward to the commit message :-)
More information about the coreboot